Reviews land on the website, Google, Amazon and eBay, while return reasons sit in a free-text field of the SAP Business One return document that nobody reads. AI classifies everything by topic and product, and managers see every week what keeps recurring and decide on the range themselves.
Built for ISO 27001NIS2 DORA
A shop receives hundreds of reviews and returns every month, but they are scattered: the WooCommerce reviews module, the Google Business Profile, ratings on the Amazon and eBay marketplaces, Facebook and Instagram comments, and the return reason that a cashier types as free text into the SAP Business One return document. A manager looks at the few worst ratings on Amazon because they affect the sales ranking, while systemic problems, for example a specific model that everyone returns because it runs small, are noticed only after the season, when the stock has already been bought again.
We bring review sources and return reasons into one flow: WooCommerce reviews through the API, the Google profile, Amazon and eBay ratings through the seller portal export, Facebook comments, and SAP Business One return documents through a periodic export. Claude or ChatGPT via API assigns every record to a topic (size, quality, delivery, mismatch with the description, service, price), determines the sentiment and links it to a specific product by the SAP Business One item code or the WooCommerce order. The results are written to a data table, from which a Power BI report shows the topic distribution, the products with the most recurring problems and the change from week to week. Every Monday managers receive a short summary with concrete examples and links to the original entries. Decisions about the range, descriptions or the supplier are made by people, and the system only shows where to look. The model runs on the Anthropic API, Amazon Bedrock in an EU region or the OpenAI platform without training on your data, and customer names are removed before analysis.
Four steps from input to a record in your system, with human approval wherever there is risk.
Every day the flow picks up new reviews from WooCommerce, the Google profile, the Amazon and eBay seller portals and Facebook, and return records with the reason field from the SAP Business One return document export. Customer names and contacts are removed before the text is passed to the model.
For each record the model assigns one or more topics from your list, determines the sentiment and links it to a product by the SAP Business One item code, the WooCommerce order number or the name. Unclear records are flagged so that a person can clarify the category.
In the Power BI report you see the topic distribution by source (Amazon, eBay, website, returns), the products with the most problems and the change from week to week. On Monday managers receive a summary with the three most frequent problems and examples, and they make the decisions themselves.
Every month we review the records the model could not assign and extend the topic list or the product-linking rules. New sources, for example another marketplace, are added without rebuilding the report.
Collecting reviews from WooCommerce, the Google profile, Amazon, eBay and Facebook, and return reasons from the SAP Business One return document export.
A topic list agreed with you and product-linking logic based on SAP Business One item codes, adapted to your categories and return forms.
A report with topics, sentiment, products, sources and the weekly change, available to managers and the merchandising team.
A Monday summary by email or in Teams with examples, and our support when the Amazon or eBay export formats change.
Claude or ChatGPT via API + review collection from WooCommerce, Google, Amazon and eBay, return reasons from a SAP Business One export, Power BI report. We connect through an API, a file export or an email flow, depending on your version. Read more about this service and its limits on the service page.
This solution is for analysis, not for replies. If you want, we can add draft replies to Google and marketplace reviews as a separate stage, but a person always publishes publicly.
We take Amazon and eBay ratings from the seller portal export or through the marketplace interface if your account has one, and SAP Business One return documents through a periodic export or the API, depending on the version. We choose the integration method after a check, and we do not claim an official partnership with these platforms.
In our experience it becomes meaningful from a few dozen records per week, because recurring problems show up quickly. For smaller volumes we suggest a monthly summary instead of a weekly one.
A new cashier at the Lightspeed till does not know whether a return without a receipt is allowed, what promotion runs today or what to do with a…
A supplier invoice, a delivery note, the SAP Business One order and the warehouse receipt rarely match, and the differences surface only after…
Customers ask whether an item is in the store, when an order will arrive and how to return it, while the support team checks the same things in…
A new supplier sends a thousand products with a code, a price and a few photos, and someone has to write the descriptions, categories and…
This service is delivered by specialists with valid international certifications. Their expertise is confirmed by independent organisations, not by us.
AI systems are deployed with protection against data leakage and model abuse.
The AI security programme and its controls are managed following the ISACA methodology.
AI risks are identified and managed like any other IT risk.
We reply within 1 business day
